Abstract-Particle accelerators are host to myriad nonlinear and complex physical phenomena. They often involve a multitude of interacting systems, are subject to tight performance demands, and should be able to run for extended periods of time with minimal interruptions. Often times, traditional control techniques cannot fully meet these requirements. One promising avenue is to introduce machine learning and sophisticated control techniques inspired by artificial intelligence, particularly in light of recent theoretical and practical advances in these fields. Within machine learning and artificial intelligence, neural networks are particularly well-suited to modeling, control, and diagnostic analysis of complex, nonlinear, and time-varying systems, as well as systems with large parameter spaces. Consequently, the use of neural network-based modeling and control techniques could be of significant benefit to particle accelerators. For the same reasons, particle accelerators are also ideal test-beds for these techniques. Many early attempts to apply neural networks to particle accelerators yielded mixed results, due to the relative immaturity of the technology for such tasks. The purpose of this paper is to re-introduce neural networks to the particle accelerator community and report on some work in neural network control that is being conducted as part of a dedicated collaboration between Fermilab and Colorado State University (CSU). We describe some of the challenges of particle accelerator control, highlight recent advances in neural network techniques, discuss some promising avenues for incorporating neural networks into particle accelerator control systems, and describe a neural network-based control system that is being developed for resonance control of an RF electron gun at the Fermilab Accelerator Science and Technology (FAST) facility, including initial experimental results from a benchmark controller.

Many modern and future particle accelerators employ high gradient superconducting RF (SRF) to generate beams of high energy, high intensity and high brightness for research in high energy and nuclear physics, basic energy sciences, etc. In this paper we report the record performance large-scale SRF system with average beam accelerating gradient matching the International Linear Collider (ILC) specification of 31.5 MV m −1 . Design of the eight cavity 1.3 GHz SRF cryomodule, its performance without the beam and results of the system commissioning with high intensity electron beam at Fermilab Accelerator Science and Technology (FAST) facility are presented. We also briefly discuss opportunities for further beam studies and tests at FAST including those on even higher gradient and more efficient SRF acceleration, as well as exploration of the system performance with full ILC-type beam specifications.

The Main Injector (MI) at Fermilab is planning to use multi-batch slip stacking scheme in order to increase the proton intensity at the NuMI target by about a factor of 1.5. [1] [2] By using multi-batch slip stacking, a total of 11 Booster batches are merged into 6, 5 double ones and one single. We have successfully demonstrated the multibatch slip stacking in MI and accelerated a record intensity of 4.6E13 particle per cycle to 120 GeV. The technical issues and beam loss mechanisms for multibatch slip stacking scheme are discussed.

A concept of a high-power magnetron transmitter utilizing the vector addition of signals of Continuous Wave (CW) magnetrons, injection-locked by phase-modulated signals, and intended to operate within a wideband control feedback loop in phase and amplitude, is presented. This transmitter is proposed to drive Superconducting RF (SRF) cavities for intensity-frontier GeV-scale proton/ion linacs, including linacs for Accelerator Driven System (ADS). The transmitter performance was verified in experiments with CW, S-Band, 1 kW magnetrons. A wideband dynamic control of magnetrons, required for the superconducting linacs, was realized using the magnetrons, injectionlocked by the phase-modulated signals. The capabilities of the magnetrons injection-locked by the phase-modulated signals and adequateness for feeding of SRF cavities have been verified by measurements of the transfer function magnitude characteristics of single and 2-cascade magnetrons in the phase modulation domain, by measurements of the magnetrons phase performance and by measurements of spectra of the carrier frequency of the magnetrons. At the ratio of power of locking signal to output power of ≥ -13 dB (in 2-cascade scheme per magnetron, respectively) we demonstrated a phase modulation bandwidth of over 1.0 MHz for injection-locked CW single magnetrons and a 2-cascade setup, respectively. The carrier frequency spectrum (width of ~ 1 Hz at the level of <-60 dBc) of the magnetron, injection-locked by a phase-modulated signal, did not demonstrate broadening at wide range of magnitude and frequency of the phase modulation. The wideband dynamic control of output power of the transmitter model has been first experimentally demonstrated using two CW magnetrons, combined in power and injection-locked by the phasemodulated signals. The experiments with the injection-locked magnetrons adequately emulated the wideband dynamic control with a feedback control system, which will allow to suppress parasitic modulation of the accelerating field in the SRF cavities, resulted from mechanical noises, phase perturbations, caused by cavity beam loading and cavity dynamic tuning errors, low-frequency ripples of the magnetron power supplies, etc. The magnetron transmitter concept, tests of the transmitter models and injection-locking of magnetrons by phase-modulated signals are discussed in this work.

No abstract